HUMAN AND ECONOMIC BURDEN OF GENERALIZED ANXIETY DISORDER

Transcription

1 DEPRESSION AND ANXIETY 25:72 90 (2008) Research Review HUMAN AND ECONOMIC BURDEN OF GENERALIZED ANXIETY DISORDER Deborah L. Hoffman, Ph.D., 1 Ellen M. Dukes, Ph.D., 2 and Hans-Ulrich Wittchen, Ph.D. 3 The goal of the current work is to provide a comprehensive review and interpretation of the literature on the human and economic burden of generalized anxiety disorder (GAD) and how it compares with that of other mental disorders. The term human burden is used to describe quantified impairments in role functioning and quality of life (QOL). Economic burden describes costs related to health care resource utilization and lost work. A review of 34 studies reporting original quantitative data on associations between GAD and role functioning, QOL, and/or economic costs was undertaken. GAD was defined by DMS-III-R, DSM-IV, or ICD-10 DCR. Persons with GAD (both with and without a comorbid mental disorder) described significant impairments due to both physical and emotional problems. Studies typically showed that role and QOL impairments of GAD were at least comparable in magnitude to those of other anxiety disorders, somatoform disorders, and physical conditions, and greater than those of substance use disorders. Large representative studies showed that role impairments of pure GAD were similar in magnitude to those of pure MDD. Studies of DSM-IV disorders showed that QOL impairments of GAD were at least comparable in magnitude to those of MDD; studies of DSM-III-R disorders showed the opposite pattern. GAD was associated with considerable economic costs owing to lost work productivity and high medical resource use. Quality of care initiatives that have been implemented to increase recognition and improve treatment outcomes for persons with MDD should be extended to the effective management of GAD. Depression Anxiety 25:72 90, & 2006 Wiley-Liss, Inc. Key words: anxiety disorders; quality of life; disability evaluation; costs and cost analysis; economics; medical; review This literature review examines aspects of the human and economic burden of generalized anxiety disorder (GAD). The first objective was to consolidate what is known about the independent associations between GAD and impairments in role functioning and quality of life (QOL). The terms role functioning and QOL together encompass concepts of social, family, and occupational functioning, perceived emotional and physical health status, well-being and satisfaction with aspects of daily life; impairments in these areas are referred to as human burden in the current review. The second objective was to describe the economic costs of GAD related to use of health care resources and lost work. The third objective was to compare how the human and economic burden of GAD compares with that of other mental disorders. GAD is characterized by persistent worry and a number of mental and somatic symptoms that occur more days than not for a period of at least 6 months 1 Independent Outcomes Research Consultant, New Haven, Connecticut 2 Pfizer, Inc., New York, New York 3 Institute for Clinical Psychology and Psychotherapy, Dresden University of Technology, Dresden, Germany Correspondence to: Deborah Hoffman, 436 Orange St., New Haven CT dhoffmanphd@comcast.net. Received for publication 9 May 2006; Accepted 7 June 2006 DOI /da Published online 4 December 2006 in Wiley InterScience (www. interscience.wiley.com). r 2006 Wiley-Liss, Inc.

2 Research Review: Burden of GA 73 [APA, 1980, 1994; WHO, 1993]. Persons with GAD feel worried, tense, or apprehensive most of the time about everyday events and activities [APA, 1980, 1994; WHO, 1993]. Common symptoms can include restlessness and inability to relax, fatigue, sleep disturbance, irritability, muscle tension, muscle aches, headache, chest pain, and other types of pain [APA, 1980, 1994; WHO, 1993]. Persons with GAD may describe or emphasize either the worry or the somatic effects of chronic tension [Ballenger et al., 2001]. Consequently, the clinical presentation of GAD can differ markedly, depending on whether patients emphasize mental or somatic anxiety symptoms [Ballenger et al., 2001]. GAD affects 5.7% of the population at some point in life [Kessler et al., 2005a] and is the most common anxiety disorder in primary care [Ormel et al., 1994; Wittchen et al., 2002]. The majority of persons with GAD experience at least one other mental disorder at some point, most commonly major depressive disorder (MDD) [Massion et al., 1993; Stein, 2001; Wittchen et al., 1994]. Persons with GAD also frequently experience comorbid physical disorders (e.g., peptic ulcer disease, diabetes, and irritable bowel syndrome), as well as cardiac and other somatic symptoms that often have no identifiable physiologic etiology [Carter and Maddock, 1992; Goodwin and Stein, 2002; Grigsby et al., 2002; Kane et al., 1988; Louge et al., 1993; Lydiard, 2000; Lydiard et al., 1993; Pearce et al., 1990; Roy-Byrne and Wagner, 2004]. If left untreated, GAD tends to follow a chronic course with periodic exacerbations [Ballenger et al., 2001; Wittchen et al., 2003]. Typically, persons with GAD experience symptoms for 5 10 years before receiving a diagnosis and treatment [Ballenger et al., 2001]. The significance of GAD has been a source of debate [Breier et al., 1985; Kessler, 2000; Kessler et al., 1999, 2002; Massion et al., 1993; Olfson et al., 1997]. An issue central to this debate is whether or not GAD is uniquely associated with impaired role functioning and QOL. A primary research question has been whether impairments of GAD are due to the disorder itself as opposed to MDD and/or other disorders that commonly co-occur with it. A related research question has been whether persons with GAD are substantially more impaired than persons without the disorder. If so, empirical findings would support the importance of recognizing and treating GAD, whether or not it is part of a comorbid condition. In consolidating information about the burden of GAD, it is important to describe how GAD-related impairments compare with those of other mental disorders. It is also important to describe economic costs associated with GAD sufferers use of health care services and lost work. Information about how GAD contributes to the overall human and economic burden of mental disorders can assist decision makers in determining how to allocate scarce health care resources and how to prioritize treatment of GAD in their health care agendas. For example, recognition of MDD as a particularly burdensome disorder has led to the development of targeted educational campaigns and disease management programs. These initiatives appear to be associated with positive clinical outcomes [Greenfield et al., 2000; Ofman et al., 2004; Olfson et al., 2002; Wang et al., 2005]. This literature review consolidates quantitative evidence on the human and economic burden of GAD by addressing three questions: What is known about associations between GAD and aspects of role functioning and QOL? What is known about the economic costs of GAD? How does the burden of GAD compare with that of other disorders? MATERIALS AND METHODS A literature search was performed in the MEDLINE and PsychLIT electronic databases for the reference period 1990 to Search terms included: generalized or generalised anxiety disorder and burden, disability, impairment, quality of life, work, utilization, primary care, specialties medical, economic and cost. Reference lists of published articles were also handsearched. Peer-reviewed studies published in English were included in the current review if they reported original, quantitative data on associations between GAD and role functioning, QOL, and/or economic costs due to use of health care resources or lost work productivity. Clinical trials of GAD were included if they assessed role functioning or QOL with a validated measure. Studies examining the cost-effectiveness of alternative treatments for GAD were not considered. Studies must have defined GAD using one of three diagnostic classification systems that require a 6-month symptom duration to diagnose the disorder: the DSM- III-R [APA, 1980], DSM-IV [APA, 1994], or ICD-10 Diagnostic Criteria for Research [WHO, 1993]. ICD- 10 DCR is hereafter referred to as ICD-10 for parsimony. ICD-10 places a greater emphasis on somatic symptoms of GAD than the DSM systems. This review summarizes data from 34 studies that satisfied the inclusion criteria described above (Table 1). Of these, 31 studies measured role functioning and/or QOL among persons with GAD in the general community (n 5 12), primary care (n 5 10), and mental health settings (n 5 9) (Table 1). The majority (n 5 9 of 15) of studies based QOL assessment on the MOS 36-item short-form health survey (SF-36) [Ware and Sherbourne, 1992] (Table 1). Table 2 describes the SF- 36. Three studies (n 5 3) examined the economic burden of GAD related to use of health care resources and lost work productivity/absenteeism (Table 1). Studies on GAD often examined at least one other disorder, particularly MDD. This review describes how findings on the burden of GAD compare with those of the other disorder(s). A variety of methods were used to examine how GAD (and the other disorder[s]) were associated with

3 74 Hoffman et al. TABLE 1. Summary of diagnostic classification system and outcome(s) assessed in thirty-four studies examining role functioning, QOL, and/or economic costs of GAD Role functioning a Setting reference Classification system Impairment days d Social Family Work Overall Physical QOL b Cost c General population Judd et al., 1998 DSM-III-R X Wittchen et al., 1994 DSM-III-R X Kessler et al., 1999 DSM-III-R X X Kessler et al., 2002 DSM-III-R X Stein et al., 2004 DSM-III-R X Cramer et al., 2005 DSM-III-R X Bijl and Vaveli, 2000 DSM-III-R X e Wittchen et al., 2000 DSM-III-R X X X e Sanderson and Andrews, 2002 DSM-IV X e Hunt et al., 2004 DSM-IV X X X e Andrews et al., 2000 ICD-10 X e Kessler et al., 2005 DSM-IV X Greenberg et al., 1999 DSM-III-R X Andlin-Sobocki et al., 2005 DSM-IV X Primary care Olfson et al., 1997 DSM-IV X X X Olfson et al., 2000 DSM-IV X X X Wittchen et al., 2002 DSM-IV X Ormel et al., 1994 ICD-10 X X X Weiller et al., 1998 ICD-10 X X Maier et al., 2000 ICD-10 X Schonfeld et al., 1997 DSM-III-R X e Spitzer et al., 1995 DSM-III-R/ X e DSM-IV Jones et al., 2001 DSM-IV X e Mental health Massion et al., 1993 DSM-III-R X Kennedy et al., 2002 DSM-III-R X X X Mavissakalian and Zamar, 2000 DSM-III-R X X X Wetherell et al., 2004 DSM-IV X e Diefenbach et al., 2003 DSM-IV X Souetre et al., 1994 DSM-III-R X Clinical Trials Pollack et al., 2001 DSM-IV X Rosenthal, 2003 DSM-IV X Rickels et al., 2003 DSM-IV X Allgulander et al., 2004 DSM-IV X X Stanley and Beck, 2003 DSM-IV X a Role functioning was assessed using instruments that measured: social, family and occupational and/or overall role functioning. Role functioning outcomes were frequently based on the Sheehan Disability Scale (SDS; Sheehan et al., 1996). The SDS assesses occupational, social, and family role functioning with three single-item scales. SDS scores can range from 0 to 10, with higher scores indicating worse role functioning (Sheehan, 2000). SDS scores can be summed in to a measure of overall role impariment. b QOL was assessed using instruments that measured a person s subjective perceptions about one or more of the following areas: physical health status, mental health status, general well-being, and satisfaction with various aspects of daily life. c Costs included direct costs related to use of health care resources and indirect costs related to lost work productivity/absenteeism. d The number of days in the past month in which normal activities could not be performed or were limited because of symptoms. e QOL assessment was commonly based on the SF-36 (Ware and Sherburne, 1992) or one of its briefer counterparts (Stewart et al., 1988; Ware et al., 1996). role functioning, QOL, and economic outcomes. One common method was to examine disorder-specific outcomes in different subgroups. Common diagnostic subgroups comprised persons with a pure, single, or comorbid disorder. Table 3 shows how these subgroups were defined and are used to describe the burden of GAD in this review. Reporting outcomes associated with a pure or single disorder separately from those with a comorbid disorder was one method used to examine whether or not a particular disorder was uniquely impairing (i.e., whether or not impairments were due to the disorder itself as opposed

4 Research Review: Burden of GA 75 TABLE 2. Description of the Medical Outcomes Study 36- item Short Form Health Survey (SF-36) a Scale Name Physical functioning Role-physical General health Bodily pain Mental health Vitality Role-emotional Social functioning Physical component summary (PCS) b Mental component summary (PCS) b Description b Limitations in physical activities because of health problems Limitations in usual role activities because of physical health problems Perceived general health status Amount of bodily pain and any limitations that result from it Psychological distress and well-being Energy and fatigue Limitations in usual role activities because of emotional difficulties Limitations in social activities because of physical or emotional difficulties An empirically derived summary scale based on scores from the four scales assessing physical domains An empirically derived summary scale based on scores from the four scales assessing mental health domains a Ware and Sherbourne, 1992 p b Scale scores range from 0 100; higher scores represent better functioning. c Summary scores have a mean of 50 and a standard deviation of 10 in a representative US population in Summary scales were not consistently used in analysis of SF-36 data. TABLE 3. Description of diagnostic subgroups used to examine outcomes Diagnostic subgroup Single GAD Pure GAD Comorbid GAD Pure MDD to the disorder(s) that co-occurred with it). Several studies compared impairments among persons with GAD and subthreshold GAD. Persons with subthreshold GAD either do not meet the temporal or severity components required for diagnosis. Descriptions of impairments among those with subthreshold GAD are not included in this review. RESULTS Description GAD with no other comorbid disorder under study GAD with no MDD GAD with at least one other comorbid mental disorder under study MDD with no GAD GAD, generalized anxiety disorder; MDD major depressive disorder. GENERAL POPULATION STUDIES EXAMINING ROLE AND/OR QOL IMPAIRMENTS OF GAD A total of 12 studies examined role functioning and/ or QOL among persons with GAD in the general population (Table 1). These studies used nationally representative data from Australia [Andrews et al., 2001], Canada [Offord et al., 1996], Germany [Wittchen et al., 1998], the Netherlands [Bijl et al., 1998], Norway (National Population Register in Oslo), and the US [Kessler et al., 1994, 1997]. The extent to which role impairments of GAD were independent from MDD was examined in a nationally representative US sample [Kessler et al., 1999, 2002; Wittchen et al., 1994]. Analyses were based on current, past-year, and lifetime DSM-III-R disorders. Respondents with comorbid GAD/MDD were more likely than other respondents to describe impairments in role functioning. However, respondents with pure GAD and pure MDD were more likely to report role impairments than those with neither disorder [Kessler et al., 1999, 2002; Wittchen et al., 1994]. For example, a study of current disorders showed that 63% of respondents with comorbid GAD/MDD (n 5 35) reported at least one impairment day in the past month compared with 5.9% of respondents with neither disorder (n ); the corresponding proportion (35%) was the same for respondents with pure GAD (n 5 52) and pure MDD (n 5 232) [Kessler et al., 2002]. A study of past-year DSM-III-R disorders showed that 27% of respondents with comorbid GAD/MDD (n 5 99) reported the highest level of social impairment on a measure of social functioning compared with 8.4% of respondents with neither disorder (n ); the corresponding proportion was 24% and 23% for respondents with pure GAD (n 5 92) and pure MDD (n 5 489), respectively [Kessler et al., 1999]. A similar pattern was observed for respondents with lifetime GAD when comorbidity with other affective disorders was examined [Judd et al., 1998]. Studies of both current and past-year disorders showed that role impairments of pure GAD were comparable in magnitude to those of pure MDD [Kessler et al., 1999, 2002]. Findings were maintained after controlling for the effects that might be due to other mental disorders and sociodemographic factors [Kessler et al., 1999, 2002]. To illustrate, findings for current disorders are described [Kessler

5 76 Hoffman et al. et al., 2002]. Pure GAD was associated with an average of between 1.5 and 5.4 impairment days in the past month, after adjusting for the presence of other mental disorders and sociodemographic characteristics. The independent incremental effects of GAD among persons who also had MDD were similar in magnitude (2.6 to 4.7 days) to those of MDD in persons who also had GAD. Past-month s role impairment of GAD was in the range previously observed for chronic medical conditions such as ulcers, arthritis, diabetes, and autoimmune disease [Kessler et al., 2001]. Moreover, in persons with comorbid GAD/MDD, GAD significantly contributed to role impairments over and above that which could be explained by MDD alone. The extent to which QOL impairments of DSM- III-R GAD were independent of DSM-III-R MDD (and vice versa) was examined in a nationally representative Canadian sample [Stein and Heimberg, 2004]. This study examined associations between QOL, GAD, and MDD, adjusting for each along with lifetime dysthymia, age, gender, and social class. Disorders were based on DSM-III-R criteria. QOL outcomes were based on the Well-being scale [McDowell and Newell, 1987] and a series of five questions about satisfaction in different life domains (main activity, family relationships, friendships, leisure activities, and income). Well-being outcomes were examined from both a past-year and lifetime perspective. Dissatisfaction outcomes were examined exclusively from a lifetime perspective. GAD respondents were significantly more likely than those without the disorder to report poor overall well-being, both in the past year and at some time in life (adjusted odds ratio [OR] 4.86, 95% confidence interval [CI]: ; OR 4.06, 95% CI: , respectively). GAD respondents were also significantly more likely than those without the disorder to report dissatisfaction with one s main activity and one s family life (adjusted OR 5.15, 95% CI: ; OR 5.64, 95% CI: , respectively). MDD respondents were significantly more likely than those without the disorder to report poor well-being and dissatisfaction on all outcomes considered (adjusted ORs ranged from 2.01 [dissatisfaction with income] to [overall well-being in the past year]). Findings for past-year GAD in the current study were similar to those found for GAD in a nationally representative Norwegian sample [Cramer et al., 2005]. Associations between 15 mental disorders and QOL and role functioning outcomes were examined in a nationally representative Dutch sample (n ) [Bijl and Ravelli, 2000]. Analyses were based on DSM-III-R disorders in the past year. QOL outcomes were based on the SF-36 (Table 2). Respondents with single GAD (n 5 31) reported mean SF-36 scores ranging from 58.9 (Vitality) to 90.8 (Physical Functioning). Compared with respondents with no mental disorder (n ), single GAD respondents had significantly lower mean scores on two scales: Vitality (73.8 vs. 58.9) and Role- Emotional (95.4 vs. 76.5) (P so.001). Associations between GAD and impaired vitality and emotional role functioning were maintained after controlling for sociodemographic characteristics, history of abuse or neglect before age 16, and the presence of a somatic disorder (adjusted bs of 0.16 and 0.21 for Vitality and Role Emotional, respectively; P so.001). Respondents with single GAD showed a nonsignificant trend toward a greater number of days in the past month when normal activities were limited because of symptoms than respondents with no mental disorder (M and 1.1, respectively). However, GAD significantly predicted disability days after adjusting for the potential confounding characteristics described above (adjusted b , Po.001). Of the disorders under study, GAD, MDD, and dysthymia showed the strongest independent associations with Vitality, Role-Emotional, and disability days (adjusted bs ranging from 0.16 [GAD and Vitality] to 0.22 [disability days and all three single disorders], P so.001). Overall, GAD was less impairing than mood disorders. In contrast, GAD was more impairing than other anxiety disorders (agoraphobia, social phobia, simple phobia and obsessive-compulsive disorder [OCD], and panic disorder [PD]), substance use disorders (alcohol/drug abuse and dependence), schizophrenia, and bulimia nervosa. Investigators noted that findings for schizophrenia and bulimia nervosa should be interpreted cautiously because there were very few cases (single schizophrenia [n 5 4] and single bulimia nervosa [n 5 8]; n s were not reported for total cases within the particular disorder groups). The differential impact of GAD and/or MDD on role and QOL outcomes was examined in a nationally representative German sample (n ) [Wittchen et al., 2000]. Analysis was based on DSM-IV disorders in the past year. As in previous studies [Kessler et al., 1999, 2002; Wittchen et al., 1994], impairments were examined among four mutually exclusive groups: pure GAD, pure MDD, comorbid GAD/MDD, and neither GAD nor MDD. Role impairment outcomes are presented in Table 4. Across all measures, comorbid GAD/MDD respondents were more likely than other respondents to report impaired role functioning. Notably, pure GAD respondents were more likely to report impaired role functioning than pure MDD respondents. QOL outcomes assessed by the SF-36 were also reported separately for the four study groups (Table 5). Respondents with GAD and/or MDD reported significantly lower mean scores on all eight SF-36 scales compared with those with neither disorder. In general, GAD was associated with worse QOL outcomes than MDD. Pure GAD respondents showed significantly lower scores on General Health, Mental Health, Role-Emotional, and Vitality scales than respondents with pure MDD. Comorbid GAD/MDD respondents showed significantly lower scores on General Health, Mental Health, Social Functioning, and Vitality scales

6 Research Review: Burden of GA 77 TABLE 4. Outcomes of the impairment measures for 12-month generalized anxiety disorder (GAD) and major depressive disorder (MDD) in a representative German sample a Pure GAD N 5 33 Pure MDD N GAD/MDD N 5 40 Neither GAD nor MDD N Impairment measure N %W 95% CI N %W 95% CI N %W 95% CI N %W 95% CI Self-perceived health Excellent or very good Good Fair Past month days impaired or more Past month s overall activity reduction b % % % % Z50% a Data are from the German National Health Interview and Examination Survey-Mental Health Supplement (GHS) (Wittchen et al., 1989). b Based on number of days limited or lost due to symptoms and calculated as a past month overall work productivity reduction percentile. Wittchen HU, Carter RM, Pfister H, Montgomery SA, Kessler RC Disabilities and quality of life in pure and comorbid generalized anxiety disorder and major depression in a national survey. International Clinical Pharmacology 15: Copyright 2000 Lippincott, Williams and Wilkins. Adapted by permission. TABLE 5. SF-36 scores associated with 12-month generalized anxiety disorder (GAD) and/or major depressive disorder (MDD) in a nationally representative German sample a Pure GAD N 5 33 Pure MDD N GAD/MDD N 5 40 Neither GAD nor MDD N SF-36 scale Mean 95% CI Mean 95% CI Mean 95% CI Mean 95% CI General b,c b,d b health Physical b b b functioning Role physical b b b Bodily pain b b b Mental health b,c b,d b Social b b,d b functioning Role b,c b b emotional Vitality b,c b,d b PCS e MCS e b,c b,d b a Data are from the GHS. Some scores contain missing data. All scores, except the sum scores, are standardized to 0-100; higher scores indicate better QOL. b Mean is significantly different from the mean in the neither GAD nor MCD group ; Po0.05. c Mean is significantly different from the mean in the pure MDD group; Po0.05. d Mean is significantly different from the mean in the GAD and MCD group; Po0.05. e Summary scores are z-transformed using a standardized US population, i.e., mean of 50 and a standard deviation of 10. Wittchen HU, Carter RM, Pfister H, Montgomery SA, Kessler RC Disabilities and quality of life in pure and comorbid generalized anxiety disorder and major depression in a national survey. International Clinical Pharmacology 15: Copyright 2000 Lippincott, Williams and Wilkins. Adapted by permission.

7 78 Hoffman et al. than pure MDD respondents. No significant differences in SF-36 scale scores were observed between respondents with comorbid GAD/MDD and pure GAD. Respondents with GAD and/or MDD reported SF-36 Physical Component Summary (PCS) scores that were similar to those of respondents with neither disorder. In contrast, respondents with GAD and/or MDD showed significantly lower mean SF-36 Mental Component Summary (MCS) scores than those with neither disorder. Associations between GAD and/or MDD and risks of impairment were examined further in analyses controlling for sociodemographics and psychiatric comorbidity (Table 6). Compared with respondents who had neither GAD nor MDD, respondents with GAD and/or MDD showed significantly increased risks for reporting fair or poor health, at least 6 impairment days in the past month, an overall activity reduction of at least 30%, and an impaired SF-36 MCS score (i.e., a score r40). In contrast, respondents with GAD and/or MDD were no more likely than those with neither disorder to show an increased risk of impairment on the SF-36 physical summary scale. Significant differences were not observed between any associations for pure GAD, pure MDD, and comorbid GAD/MDD. The effect of GAD on a summary measure of mental health-related QOL (SF-12) [Ware et al., 1996] was examined in a nationally representative Australian sample [Andrews et al., 2000; Hunt et al., 2004; Sanderson and Andrews, 2002]. Mean SF-12 MCS scores were reported for respondents with current DSM-IV GAD and current ICD-10 GAD, both with and without psychiatric comorbidity (Table 7). These SF-12 MCS scores fell more than one standard deviation below the population average of 50. Mean SF-12 MCS scores in the present study were similar to mean SF-36 MCS scores reported by German respondents with pure GAD (M ) and comorbid GAD/MDD (M ) [Wittchen et al., 2000; see Table 5 in this article]. Mean SF-12 MCS scores reported by respondents with DSM-IV GAD were compared with those reported by respondents with other DSM-IV mental disorders. Higher SF-12 MCS scores indicate better mental health-related QOL. Single GAD respondents had a higher mean SF-12 MCS score than single MDD respondents (single MDD: n 5 152, M [SE ]; t 5 2.1, Po.05) [Hunt et al., 2004]. Single GAD respondents did not differ from single PD respondents with respect to mean SF-12 MCS scores (single PD: n 5 25, M [SE 5 5.0]; (t 5 1.1, P 5.28) [Hunt et al., 2004]. However, when associations were examined in multivariate analyses, impairment of GAD was comparable in magnitude to that of MDD, and greater in magnitude than that of PD [Sanderson and Andrews, 2002]. The extent to which 14 different DSM-IV disorders uniquely impacted SF-12 MCS scores was examined in multivariate regression analysis [Sanderson and Andrews, 2002]. Regression analysis was used to predict differences in SF-12 MCS scores between respondents with a particular disorder and those who did not have the disorder. Predicted difference scores were adjusted for sociodemographic characteristics, the presence of a physical disorder, and psychiatric comorbidity. By comparing predicted difference scores associated with each disorder, it was possible to compare how SF-12 MCS outcomes for GAD compared with those of the other 13 disorders under study. GAD and mild, moderate, and severe depression were the only disorders to demonstrate a strong independent association with SF-12 MCS impairment (predicted difference scores compared with those without the disorder were [SE ], [SE ], [SE ], and [SE ] for GAD, TABLE 6. Conditional associations of generalized anxiety disorder (GAD) and major depressive disorder (MDD) with health status and psychosocial impairments in a representative German sample a Pure GAD N 5 33 Pure MDD N GAD/MDD N 5 40 OR 95% CI OR 95% CI OR 95% CI Perceived health: fair or poor Past month s days impaired: 6 or more Overall activity reduction Z30% SF 36 PCS score r40 b SF-36 MCS score r a Data are from the GHS; OR 5 odds ratio; comparison group: no 12-month GAD and no 12-month MDD. Controlled for age, gender and other psychopathology. OR, odds ratio; CI, confidence interval. b r40 is worse than approximately 84% of a representative US population. Wittchen HU, Carter RM, Pfister H, Montgomery SA, Kessler RC Disabilities and quality of life in pure and comorbid generalized anxiety disorder and major depression in a national survey. International Clinical Pharmacology 15: Copyright 2000 Lippincott, Williams and Wilkins. Adapted by permission.

8 Research Review: Burden of GA 79 TABLE 7. Mean SF-12 MCS scores for respondents with current generalized anxiety disorder (GAD), with and without psychiatric comorbidity a Reference GAD Sample N Mean SF-12 MCS Score Sanderson and Andrews, 2002 DSM-IV (0.90) b Hunt et al., 2004 DSM-IV single GAD (1.5) b Andrews et al., 2002 ICD-10 GAD (10.9) c Andrews et al., 2002 ICD-10 Core GAD d (10.2) c a Data source: Australian National Survey of Mental Health and Wellbeing (Andrews et al., 2001). b (SE). c (SD). d GAD identified as the main or only complaint in the previous one-month period. MDD, and dysthymia, respectively; P so.001). PD, agoraphobia, and alcohol dependence showed moderate independent associations with impairment (predicted difference scores of 5.02 [SE ], 3.39 [SE ], and 2.22 [SE ], respectively; P so.01). Social phobia and drug dependence showed a mild independent association with impairment (predicted difference scores of 2.62 [SE ] and 2.87 [SE ], respectively; P so.05). The Australian data were also used to examine the extent to which GAD was uniquely associated with role impairments [Hunt et al., 2004]. Multivariate analysis was used to examine whether GAD predicted role impairment beyond that which could be explained by MDD. GAD was significantly associated with an increased risk for having at least one role impairment day in the previous month (Wald w , Po.001). GAD was also associated with a significantly increased risk for showing moderate to severe impairment (BDQ; Wald w , Po.001) on the role functioning scale of the Brief Disability Questionnaire (BDQ) [Ormel et al., 1994; VonKorff et al., 1996]. These findings were maintained after controlling for MDD (Wald w 2 : [impaired days]; Wald w [impaired BDQ]). Additional analysis showed that single GAD respondents reported a significantly higher mean number of impairment days than the single PD respondents (single GAD: n 5 100, M days [SE 5 1.1]; single PD: n 5 25, M days [SE 5 1.0]; t 5 2.8, Po.01). Single GAD respondents did not differ significantly from single PD respondents in terms of mean impairment days (single MDD: n 5 142, M 5 9.1; t 5 1.8, P 5.07). National Comorbidity Survey-Replication (NCS-R) data were used to evaluate the relative severity of several different past-year DSM-IV mental disorders in a nationally representative US sample (n in the total part I sample) [Kessler et al., 2005b]. Severity was defined in terms of three mutually exclusive categories: mild, moderate, and serious. Severity categories were based, in part, on levels of role impairment assessed by the SDS and past-year s impairment days. Severity levels were also based on particular disorder-specific criteria. Notably, anxiety-specific criteria were not used to define the disability categories. Of the total part 1 sample (n ), 3.1% were diagnosed with GAD. Severity ratings of mild, moderate, and serious were observed in 23.1%, 44.6%, and 32.3% of those with GAD, respectively. In sum, general population studies are constrained by self-report and recall bias. However, use of general population data to examine role and QOL impairments of mental disorders offered at least two major advantages. First, analyses were based on large, representative national data sets. Second, study samples were not biased by help-seeking. Despite differences in methodologies, general population studies from around the world showed a consistent finding: GAD was uniquely associated with substantial impairments in role functioning and QOL. Although persons with GAD were more impaired when they had a comorbid disorder, persons with single and pure GAD were substantially more impaired than those without the disorder. Direct comparisons showed that role impairments of pure GAD were comparable in magnitude to those of pure MDD and in the range of those previously reported for chronic physical disorders. Impairments of GAD were at least comparable in magnitude to those of PD. Studies of DSM-IV disorders showed that QOL impairments of GAD were at least comparable in magnitude to those of MDD; studies of DSM-III-R disorders showed the opposite pattern. GAD appeared to have particularly strong negative associations with vitality, role functioning due to emotional difficulties, and overall well-being. PRIMARY CARE STUDIES EXAMINING ROLE AND/OR QOL IMPAIRMENTS OF GAD A total of nine primary care studies evaluated role functioning and/or QOL among patients with GAD (Table 1). Similar to studies of GAD in the community, primary care studies examined the extent to which impairments of GAD were independent of MDD and other mental disorders. QOL and role impairments of GAD were also compared with those of other mental and physical disorders. Role functioning was examined in primary care patients in two separate US primary care settings

9 80 Hoffman et al. [Olfson et al., 1997, 2000]. The focus of these methodologically similar studies was to examine the extent to which DSM-IV GAD and other DSM-IV disorders were uniquely associated with role impairments in work, family, and overall functioning. The first study [Olfson et al., 1997] included a total sample of 1001 randomly selected patients with scheduled primary care appointments at a health maintenance organization. Of the 37 patients diagnosed with GAD, only four had single GAD. In this study, GAD was not uniquely associated with impairments on any of the three outcome measures considered. Notably, findings were not replicated when similar methods were used in the second study that included a larger sample of patients with single GAD [Olfson et al., 2000]. The second study [Olfson et al., 2000] included a total sample of 1007 consecutive patients with scheduled appointments at an urban general medical practice. The practice was affiliated with a public hospital that serves predominantly low-income and uninsured patients. Bivariate analysis was used to compare role impairments across three groups: single GAD (n 5 34), comorbid GAD (n 5 115), and no mental disorder (n 5 697). Patients with GAD, as both a single and comorbid disorder, were significantly more likely than patients with no mental disorder to report missing at least 1 day of work in the past month due to symptoms (44.1% and 68.0% vs. 32.5%, respectively; P so.001). Patients with GAD, as both a single and comorbid disorder, were also more likely than patients with no mental disorder to report worse overall family/ social role functioning as assessed by mean 2-scale SDS scores. Two-scale SDS scores have a theoretical range between 0 20, with higher scores indicating worse overall family/social role functioning. Mean 2-scale SDS scores were 4.9 (SD 5 6.3), 10.1 (SD 5 7.5), and 1.7 (SD 5 3.8) for patients with comorbid GAD, single GAD, and no mental disorder, respectively; P so.001 (corresponding medians were 2, 10 and 0, respectively). Comorbid GAD patients, but not single GAD patients, were significantly more likely to describe family role functioning impairment than patients with no mental disorder (26.1% vs. 17.6% and 6.2%, respectively; Po.001). Independent associations between disorders and role impairments were examined further in multivariate regression analyses, controlling for sociodemographic factors and psychiatric comorbidity. GAD was independently associated with a 2.2-fold (95% CI: ) increased risk of work loss in the past month, a 3.0-fold (95% CI: ) increased risk of family role impairment, and an expected increase of 5.5 points (95% CI, ) on the 2-scale SDS. The expected increase on the 2-scale SDS for GAD was roughly equivalent to the 5.8 expected increases observed for MDD (95% CI: ) and PD (95% CI: ). The Generalized Anxiety and Depression in Primary Care (GAD-P) study was a 1-day survey of 20,000 consecutive patients and 558 primary care physicians in Germany [Wittchen et al., 2002]. This study examined associations between DSM-IV GAD and/or DSM-IV major depressive episode (MDE) and occupational role impairment. Occupational role impairment was defined as being unable to work at least 1 day in the past month because of symptoms. The study was unique in that it examined disorder-specific occupational role impairment outcomes due to physical symptoms separately from those due to psychiatric symptoms. Table 8 shows rates of occupational role impairment reported by primary care patients with GAD and/or MDE. Patients with GAD and/or MDE described significantly greater occupational role impairment due to physical and psychiatric symptoms compared with patients with neither disorder. Occupational role TABLE 8. Rates of occupational impairment in the past month reported by primary care patients with generalized anxiety disorder (GAD) and/or major depressive episode (MDE) a No GAD/No MDE Pure GAD Comorbid GAD/MDE Pure MDE Disability/role impairment % % OR (95% CI) % OR (95% CI) % OR (95% CI) Disabled because of somatic problems Disabled because of psychiatric problems ( ) ( ) ( ) ( ) ( ) ( ) a Data are from the nationally representative German primary care study Generalized Anxiety and Depression in Primary Care (GAD-P) OR, odds ratio; CI, confidence interval. Wittchen HU, Kessler RC, Beesdo K, Krause P, Höfler M, Hoyer J Generalized anxiety and depression in primary care: prevalence, recognition and management. Journal of Clinical Psychiatry 63 (Suppl. 8) Copyright Physicians Postgraduate Press. Adapted by permission.

10 Research Review: Burden of GA 81 impairment due to psychiatric symptoms was especially common. The mean number of past month s occupational role impairment days was substantial, with 9.9 for pure GAD (mean ratio [MR] compared with neither disorder of 1.4; 95% CI: ); 15.3 for pure MDE (MR 5 2.2; 95% CI: ); and 16.5 for comorbid GAD/MDE (MR 5 2.3; 95% CI: ) [Wittchen et al., 2002:29]. The World Health Organization Psychological Problems in General Health (PPGH) was a standardized evaluation of 15 primary care sites in 14 countries. The PPGH included assessments of role functioning, based on the self-report BDQ [VonKorff et al., 1996] and the interview-administered Social Disability Schedule [Wiersma et al., 1988]. PPGH studies consistently found that a substantial proportion of GAD patients, both with and without comorbidity, reported marked impairments in role functioning [Maier et al., 2000; Ormel et al., 1994; Weiller et al., 1998]. For example, descriptive analysis of data pooled across all PPGH centers showed that GAD patients were approximately 5 times more likely to describe moderate/severe occupational dysfunction and moderate/severe physical disability than psychiatrically healthy controls (Table 9) [Ormel et al., 1994]. GAD patients also reported over three times as many impairment days in the past month as controls. In order to examine unique impairments associated with each of the disorders, analyses were repeated for patients with single disorders. As shown in Table 9, impairment in all single disorder groups was somewhat attenuated after excluding patients with comorbidity. Yet patients with single disorders showed substantially poorer role impairment outcomes compared with those without a mental disorder. The proportion (54%) of persons with single GAD describing moderate/severe TABLE 9. Prevalence, psychiatric comorbidity, and disability by current ICD-10 diagnosis, pooled across 14 primary care centers in 15 countries (weighted estimates) Moderate or Severe, % Mean Occupational role dysfunction Self-reported physical disability N Disability days in past month No. of Cases, Unweighted Current ICD diagnosis, all patients Psychiatrically well patients a ICD depressive episode ICD panic disorder ICD agoraphobia ICD neurasthenia ICD hypochondriasis ICD generalized anxiety ICD alcohol dependence ICD somatization disorder Current ICD diagnosis, patients with Z2 psychiatric disorders are excluded Psychiatrically well patients a ICD depressive episode ICD panic disorder ICD agoraphobia ICD neurasthenia ICD hypochondriasis ICD generalized anxiety ICD alcohol dependence ICD somatization disorder a Well patients do not include patients with a definite or a subthreshold disorder or symptomatic patients. Ormel J, Vonkorff M, Ustun B, Pini S, Korten A, Oldehinkel T Common mental disorders and disability across cultures: results from the collaborative study on psychological problems in general health care. JAMA 272: Copyright American Medical Association. Reprinted by permission.

11 82 Hoffman et al. physical role impairment was higher than that for the other seven single disorder groups under study. The extent to which untreated anxiety disorders and/or MDD impacted QOL was evaluated in a study of primary care patients enrolled in a health maintenance organization in the US [Schonfeld et al., 1997]. A total of 6703 eligible patients completed a screening assessment that included an anxiety and depression symptom inventory and the SF-36. Patients who were identified as having clinically significant anxiety symptoms and who had not received treatment for an emotional problem in the previous 6 months were administered a structured diagnostic interview. Of these anxious patients, 319 were identified as having at least one untreated DSM-III-R disorder, including: GAD (n 5 55; n 5 14 single cases); posttraumatic stress disorder (PTSD; n 5 110; n 5 25 single cases); simple phobia (n 5 94; n 5 0 single cases); social phobia (n 5 89; n 5 23 single cases); panic/agoraphobia (n 5 100; n 5 20 single cases); OCD (n 5 23; 1 single case); and MDD (n 5 158; n 5 54). A total of 4242 eligible patients scored below anxiety cut-points and served as a nonanxious comparison group. Regression analysis was used to estimate how the presence of five single disorders affected SF-36 scores predicted for a reference group of females age with no mental disorders. For the reference group, predicted SF-36 scores were 91.9 (Physical Functioning), 83.6 (Social Functioning), 82.5 (Role-Physical), 84.0 (Role-Emotional), 75.6 (Mental Health), 59.9 (Vitality), 77.8 (Bodily Pain), and 77.5 (General Health). The presence of single GAD was estimated to significantly lower predicted scores for the reference group on six scales. Specifically, single GAD was estimated to lower predicted scores by approximately 14 points on Physical Functioning, 13 on Social Functioning, 21 on Role-Physical, 28 on Role- Emotional, and 16 on Vitality and General Health (P so ). Compared to other single disorders, single GAD was estimated to have a significant impact on more QOL domains than social phobia and significantly fewer QOL domains than PD, PTSD, and MDD (significant reductions were observed on 6, 5, 7, and 8 predicted scale scores for single disorder groups GAD, social phobia, PD, and both PTSD and single MDD, respectively). Of all of the single disorder groups, single MDD was estimated to have the worst overall QOL outcomes. However, due to sampling design all patients with MDD also had clinically significant anxiety symptoms. Consequently, the extent to which MDD itself was responsible for poor outcomes is not clear. Findings are also limited by inclusion of only 14 cases of single GAD. The study also examined how each specific disorder, when it occurred as part of a comorbid disorder, uniquely added to impairment on SF-36 scores. For these analyses, the reference group consisted of females age who had multiple disorders (which accounted for the average impact of having more than one disorder on each predicted score). When occurring as part of a comorbid condition, GAD significantly added to impairment on predicted scores for Vitality and Physical Functioning scales. The presence of GAD, as part of a comorbid condition, was estimated to lower the predicted score of 45 (Vitality) and 87.3 (Physical Functioning) by 7.9 points and 6.5 points, respectively. Of all disorders occurring as part of a comorbid condition, GAD had the greatest negative impact on Vitality and the greatest negative impact next to PTSD on Physical Functioning. The extent to which single GAD impacted SF-20 [Stewart et al., 1988] scores was examined as part of a primary care study in the US (n ) [Spitzer et al., 1995]. This study used regression analysis to show the average difference in SF-20 scores between patients with 12 different disorders and patients with no mental disorder (most of whom had a physical disorder). Difference scores were adjusted for sociodemographics, site, and physical disorders. On all six SF-20 scales, single GAD was estimated to have significantly lower adjusted mean scores than respondents with no mental disorder. For patients with no mental disorder (n 5 614), mean SF-20 scores were 77.8 (Physical Functioning), 66.2 (Bodily Pain), 84.4 (Role-Functioning), 80.8 (General Health), 67.0 (Social Functioning), and 65.3 (Mental Health). Compared with patients with no disorder, single GAD patients showed an average reduction in adjusted mean SF-20 scores of 21.8 on Physical Functioning, 24.1 on Bodily Pain, 43.6 on Role-Functioning, 28.5 on General Health, and 37.9 for Mental Health (P so.001). On all SF-20 scales, the 12 single disorder groups showed significant reductions in average adjusted mean scores compared with the reference group. Overall, single GAD was associated with adjusted mean score reductions that were typically about 30 points below that of the reference group (single GAD: n 5 70; M 5 31 for overall adjusted mean score reductions). The magnitude of impairment of single GAD was lower than the overall impairment of dysthymia (n 5 78; M 5 35); roughly comparable to the overall impairment of MDD (n 5 115, M 5 32), PD (n 5 36; M 5 31), and mutisomatoform disorder (n 5 82; M 5 30); and greater than overall impairment of anxiety disorder not otherwise specified (NOS) (n 5 90; M 5 23), minor depression (n 5 64; M 5 20), partial depression (n 5 63; M 5 21), hypochondriasis (n 5 22; M 5 25), somatoform disorder NOS (n 5 42, M 5 21), and binge eating (n 5 30, M 5 27). A longitudinal primary care study examined associations between DSM-IV mental disorder groups and 1-year QOL outcomes [Jones et al., 2001]. Patients were randomly selected from primary care clinics at a